Double returns 0

Question

Possible Duplicate:
C programming division

I am trying to convert values into the Mel Filter but, every time I output the result, it is always "0" even though it is a double. Should it be a float? Here is an article on the equation: http://en.wikipedia.org/wiki/Mel_scale

Here is the functions:

double MFCC::mel_filter(int val)
{
     double tmp;
     tmp = 2595 * log10(1 + val / 700);
     cout << tmp << endl;
     return tmp;
    }

    vector<double> MFCC::mel_frame(int size)
    {
    vector<double> mel_Frame;
    for(int i=1; (i < size); i++)
    {
          this->mel_filter(i);
    }

    return mel_Frame;
    }

In main, I will pass values such as.. (1, 2, 3, 4, 5, 6, 7 ......)

And if "1" was entered then the equation would be:

m = 2595 * log10(1 + 1 / 700) = 1.6

Any help or suggestions?

Try `1 + val/700.0` -- `/` is integer division if both operands are integer — nvlass, Dec 04 '12 at 20:55
It's worth nothing that, though you declare and return `vector mel_Frame;`, you never actual write any values into it. Your function will always return an empty vector. — user229044, Dec 04 '12 at 20:56

score 5 · Accepted Answer · answered Dec 04 '12 at 20:55

5

Your problem is at

log10(1 + val / 700)

That's using an integer division instead of a real division. Try with

log10(1 + val / 700.0)

answered Dec 04 '12 at 20:55

K-ballo

80,396
20
159
169

score 3 · Answer 2 · answered Dec 04 '12 at 20:55

3

The type of val is int. So the expression val / 700 is evaluated with integer arithmetic. The result is zero.

answered Dec 04 '12 at 20:55

Eric Postpischil

195,579
13
168
312

score 2 · Answer 3 · answered Dec 04 '12 at 20:55

2

When you use integer number literals, integer arithmetics are used. That means any decimals will be rounded down. Write the constants explicitly as decimal numbers, and double arithmetics will be used:

tmp = 2595.0 * log10(1.0 + val / 700.0);

answered Dec 04 '12 at 20:55

Philipp

67,764
9
118
153

score 2 · Answer 4 · edited Dec 05 '12 at 02:05

2

Try

log10(1 + val / (float) 700)

edited Dec 05 '12 at 02:05

thelatemail

91,185
12
128
188

answered Dec 05 '12 at 01:43

Andy White

21
2

Double returns 0

4 Answers4